Mechanical validation procedure for an adsorbent.

The mechanical validation method addresses the issue of adsorbent degradation by simulating operational scenarios and conducting breakage tests, ensuring adsorbents maintain integrity and performance in gas separation processes.

FR3142913B3Active Publication Date: 2024-11-29LAIR LIQUIDE SA POUR LETUDE & LEXPLOITATION DES PROCEDES GEORGES CLAUDE
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
FR2022012965
Authority / Receiving Office
FR · FR
Patent Type
Utility models
Current Assignee / Owner
Filing Date
2022-12-08
Publication Date
2024-11-29
Estimated Expiration
2032-12-08

AI Technical Summary

Technical Problem

Existing methods for selecting and validating adsorbents for gas separation processes fail to account for the potential mechanical degradation caused by adsorbed compounds, leading to issues such as particle breakage and dust formation, which can compromise the integrity and performance of adsorption units.

Method used

A method for mechanical validation of adsorbents involves simulating operational scenarios to determine mechanical stress, using accelerated aging techniques to replicate long-term use, and conducting breakage tests on adsorbent samples to ensure they maintain integrity under expected conditions.

Benefits of technology

Ensures that selected adsorbents can withstand the mechanical stresses of their intended use, preventing particle breakage and dust formation, thereby maintaining the integrity and performance of gas separation units over their operational lifespan.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000037_0000
    Figure 00000037_0000
  • Figure 00000038_0000
    Figure 00000038_0000
  • Figure 00000039_0000
    Figure 00000039_0000
Patent Text Reader

Abstract

Mechanical qualification method for particles of an adsorbent (66, 67) intended for a process of separating a gaseous mixture by adsorption into a most adsorbable fraction and a least adsorbable fraction, the process employing at least one adsorber (11, 16,… 60), the method comprising, in addition to steps c), d) for forming batches of particles and regenerating the adsorbent and steps f), g), h) for testing particle breakage, the following steps: i) measure the quantity in the gaseous mixture of at least one constituent whose boiling point at atmospheric pressure is above -20°C, constituents selected from water, hydrocarbons, ethers, alcohols, aldehydes, ketones, j) compare the quantity measured in step i) to a predetermined threshold value, k) if the quantity measured in step i) is above the threshold value, load the batches regenerated in step d) with the constituent(s),and submit the resulting batches to the testing, comparison and validation steps corresponding to steps f), g), h). Abbreviated figure: figure 1,
Need to check novelty before this filing date? Find Prior Art

Description

Title of the invention: Procedure for mechanical validation of an adsorbent.

[0001] The present invention relates to the mechanical validation of an adsorbent with a view to its use in a process for separation or purification by adsorption of a mixture of constituents.

[0002] Generally speaking, an adsorption process makes it possible to separate one or more molecules from a mixture containing them, by exploiting the difference in affinity of one or more adsorbents with respect to the constituent molecules of the mixture. The affinity of an adsorbent for a molecule depends on the one hand on the structure and composition of the adsorbent and on the other hand on the properties of the molecule, in particular its size, its electronic structure and its multipolar moments.

[0003] Although such a process can also be carried out in the liquid phase, we limit ourselves in the developments below to gas phase adsorption, which is currently widely used in many applications. Examples include the separation of n and iso paraffins, the separation of xylenes, alcohols, synthesis gas, the production of nitrogen or oxygen from atmospheric air, the deballasting of CO2 from combustion gases and blast furnace gases. On the purification side, there are dryers, the purification of hydrogen or helium, the purification of methane-rich gases such as natural gas or biogas, from atmospheric air before its cryogenic separation, the adsorption of trace impurities in many fluids (removal of mercury, NOx, sulfur products).

[0004] The processes using adsorption are of several types depending on whether the adsorbent is regenerable in situ or not. We can then distinguish three main families of adsorption processes: lost charge processes, TSA processes (Temperature Swing Adsorption) and finally PSA processes (Pressure Swing Adsorption).

[0005] In lost charge processes - in this case we often speak of a guard bed - a new charge is put in place when the one currently in use is saturated by impurities or more generally when it can no longer play its protective role sufficiently.

[0006] In TSA type processes, the adsorbent at the end of use is regenerated in situ, that is to say that the stopped impurities are evacuated so that the said adsorbent recovers the majority of its adsorption capacities and can start a purification cycle again, the essential regeneration effect being due to a rise in temperature. It should be noted here that the quantity of heat to be supplied and the temperature level to be used depend essentially on the quantity of impurity to be desorbed and the affinity of the adsorbent for said impurity. In the case where this affinity is high (zeolite and water system for example), it may be necessary to use temperature levels exceeding 200°C. Conversely, in the case of low affinity (activated alumina / CO2 for example) temperatures of less than 100°C may be sufficient to desorb the impurities. Regeneration often takes place at low pressure, which promotes the desorption of impurities. It is also common to push the heat front present in the adsorber at the start of the cooling phase. In these latter cases, the acronyms TPSA or TEPSA are sometimes found instead of TSA.

[0007] Finally, in PSA-type processes, the adsorbent at the end of the production phase is regenerated by the desorption of impurities obtained by means of a reduction in their partial pressure. This reduction in pressure can be obtained by a reduction in the total pressure and / or by flushing with a gas free of or containing few impurities. This gas can in certain cases be external to the unit.

[0008] It should be noted that a PSA type unit is not exempt from thermal effects. The adsorption and desorption of impurities locally create heating and cooling whose amplitude (beat) can be up to a few tens of °C when the adsorbed quantity is significant. There may also be a temperature profile between the inlet and the outlet with zones that are hotter and / or colder than the average. These thermal effects generally have a negative impact on regeneration. Pressure swing adsorption processes are used both to remove traces of impurities - for example, with a content of less than 1% in the feed gas - and to separate mixtures containing tens of % of different gases. In the first case, we generally speak of purification (for example, gas drying) and of separation in the second case (for example, production of oxygen or nitrogen from atmospheric air).In the most complex cases, we can of course have purification and separation in the same unit.

[0009] It is common to give more specific names depending on the pressure levels involved or the time required for an adsorber to return to its initial point (cycle time): - VSA processes in which adsorption takes place substantially at atmospheric pressure, preferably between 0.95 and 1.25 bar abs and the desorption pressure is lower than atmospheric pressure, typically 50 to 400 mbar abs. - MPSA or VPSA processes in which adsorption takes place at a high pressure above atmospheric pressure, typically between 1.3 and 6 bar abs, or even 10 bar abs, and desorption at a low pressure below atmospheric pressure, generally between 200 and 750 mbar abs. - The PSA processes themselves in which the high pressure is substantially higher than atmospheric pressure, typically between 3 and 50 bar abs and low pressure substantially equal to or higher than atmospheric pressure, generally between 1 and 9 bar abs. - RPSA (Rapid PSA) processes for which the pressure cycle time is typically less than one minute. - URPSA (Ultra Rapid PSA) processes for which the pressure cycle duration is of the order of a few seconds maximum.

[0010] It can also be noted that the gaseous fraction used in a PSA process can correspond to the fraction produced at high pressure but also to the fraction extracted at low pressure provided that the constituent(s) sought are the most adsorbable in the mixture.

[0011] It should be noted that these various names (TSA, TEPSA, TPS A, PSA and VS A, VPSA, MPSA, RPSA, URPSA) are not standardized and that the limits are subject to variation according to the authors. Other names exist in the literature highlighting this or that particular point of the process.

[0012] It is recalled that unless otherwise stated, the use in the description of the term PSA here covers all variants of PSA, and that by TSA we mean any type of unit in which a temperature effect is used to carry out regeneration.

[0013] It will be noted from the above paragraphs that the operating conditions of adsorption gas separation units are very varied in terms of pressure, temperature and flow rates. The rates of variation of these parameters are also very different from one case to another, particularly depending on the duration of the stages.

[0014] The adsorbents are first selected based on their physical characteristics relating to the "separation" function required by the process envisaged. Small-scale tests can be carried out based on the chosen process, but increasingly, adsorption process simulation software is used to evaluate the unit's performance based on data concerning the adsorbents. However, to be validated in the process envisaged, the designer generally follows an internal protocol in parallel which examines their compatibility with the constituents of the gas to be treated.Adsorbent / adsorbate compatibility refers to the analysis of the risks that could be caused by contact between the constituents of the gas and the adsorbent during operation, said risks being essentially linked to safety (creation of explosive zones by adsorption / desorption effect, presence of oxidant / fuel), to irreversible poisoning of the adsorbent (deposits, polymerization), to partial destruction of its internal structure (in the presence of an acidic or basic constituent), to parasitic reactions (adsorbent playing a catalyst role).

[0015] The adsorbent chosen because it leads to the best performances, while being compatible with the process, could be for example a zeolite, an activated carbon of mineral or vegetable origin, an activated alumina possibly doped, a silica gel, a carbon molecular sieve, a metallo-organic structure, an oxide or hydroxide of alkali or alkaline-earth metals, a resin, a polymer-based material, or a porous structure preferably containing a substance capable of reacting reversibly with the molecules, a substance such as amines, physical solvents, metal complexing agents, metal oxides or hydroxides for example. Zeolites for their part form a very vast family. They are made up of a set of crystals themselves formed at the base of a three-dimensional microporous skeleton of aluminosilicate of various metals.The different possible Si / Al ratios, the more or less complete cation exchanges and the substitution of other elements for the aluminium in the framework lead to hundreds of materials with different properties. We will obviously cite, without limitation, zeolites of the LSX, X, A, Y, ZSM-5, silicalite type, which are among the most used. Often, the adsorption process will use several types of adsorbents in successive layers, each of the layers being more particularly adapted to stopping one of the impurities. PSA H2 is a good example of multi-layer adsorption processes.

[0016] The most classic adsorbent materials are in the form of particles (balls, sticks, crushed materials, pellets, perforated discs, etc.) with dimensions generally ranging from 0.3 to 10 mm, more often between 0.8 and 5 mm, but also exist in more structured forms such as monoliths, wheels, parallel passage contactors, fabrics, fibers.

[0017] Adsorbents obtained by extrusion and a majority of adsorbents in the form of beads generally contain a binder which promotes their shaping. This binder generally represents 5 to 25% by weight of the particle. There are a large number of binder materials available to adsorbent suppliers: kaolinite, attapulgite, montmorillonite, sepiolite, resin, polymer, tar, silica, again in a non-exhaustive manner. After shaping the particles, post-treatment in certain cases allows the initially essentially inert binder to be transformed into an adsorbent phase. This is then incorrectly referred to as an adsorbent without binder. This is particularly the case for zeolites. It is also common to add small quantities of different components to promote shaping, to increase the number or size of pores, possibly to modify the thermal conductivity or the heat capacity.

[0018] The adsorbent mass, which can therefore comprise several types of adsorbents, generally in successive layers but sometimes in a mixture, is very largely kept immobile in one or more reactors which are given the more specific name of adsorber. It is this latter name that will be retained subsequently.

[0019] The simplest adsorber is cylindrical in shape with a vertical axis. When the flow rates to be purified become large, clusters of adsorbers of this type can be used, operating in parallel and behaving like a single adsorber.

[0020] Cylindrical adsorbers with a horizontal axis are also used.

[0021] With higher flow rates and / or if low pressure losses are sought and / or if the gas speed can be higher than the attrition speed, i.e. the speed of movement of the particles, it becomes interesting to use a radial adsorber.

[0022] For low pressure applications, substantially parallelepipedal adsorbers are also used.

[0023] Alongside these classical forms, we can find in the literature other more complex geometries intended to solve particular problems (truncated, spherical, semi-cylindrical form).

[0024] The dimensions of industrial adsorbers vary greatly, regardless of their shape. For cylindrical external envelopes, diameters ranging from approximately 0.5 m to 6 or 7 m can be found, with heights ranging from one meter to 15 m, or even 20 m and more. Similarly, in the case of parallelepiped envelopes, the sides generally have dimensions between 0.5 and 15 m.

[0025] It is appropriate that, whatever the process (guard bed, TSA, PSA), the operating conditions (pressures, temperatures, gas flows, duration of stages), the type and dimensions of the adsorber, the filling method and possibly the type of transport and the emptying method, the adsorbent retains its integrity during its period of use. By "retain its integrity", we mean both that it does not break into several pieces but also that it does not produce dust (often called fines) by premature wear of its surface. To simply define these different problems, all likely to modify the integrity of the adsorbent particle, we will speak generally of rupture of the particle, whether this concerns its external surface or its volume. In the same way, we can use the terms rupture stress or rupture test for the different phenomena cited.

[0026] As described at the beginning of this document, adsorbents are very generally in the form of particles with dimensions often ranging from 0.3 to 10 mm and mostly from 0.8 to 5 mm.

[0027] These particles will undergo shocks during the filling of the envelope which contains them (adsorber), vibrations and shocks in the event of transport and installation of the adsorber on site. During operation, they will be in contact with different fluids at variable speeds, pressures, temperatures. They will be subjected to sudden variations in pressure and temperature. Each particle will be in contact with neighboring particles, a wall, a grid, a particle being subjected to the weight of the bed, to the pressure losses of the circulating fluids, to expansions and contractions of the envelope under the effect of pressure and temperature. All these phenomena are likely to create fines, or even break the particles even if the sizing of the adsorber has been carried out according to the rules of the trade (compliance with the attrition speed for example).

[0028] It is then understood that adsorbent debris and the creation of dust can lead to local blockages at the level of filters, supports, separation fabrics between layers or even within the adsorbent mass. In addition to the increase in pressure drop through the adsorber which results in additional energy consumption, these blockages will cause local defects in the distribution of gas flows which can compromise the achievement of the expected performance. Debris and dust can have serious consequences for the equipment of the unit in question but also for neighboring units.They will be able to pass through the grids holding the adsorbent mass, block the valves, clog the instrumentation sockets and even pollute the upstream and / or downstream processes, possibly causing very serious incidents, for example in the case of machines or compact exchangers which will then be very difficult to clean due to their design.

[0029] Those skilled in the art also know that the duration of use of an adsorbent in a unit can in certain cases affect its condition and in particular reduce its capacity to withstand stresses. It is therefore necessary to test both new adsorbents and used adsorbents to avoid any incident not only at the start-up of the unit but also during operation.

[0030] From now on, we call "new" adsorbent the product as it comes out of the supplier's packaging or having been in service for only a short time, let's say of the order of one percent of its expected lifespan, for example one or two weeks, and "used" adsorbent a product, on the other hand, having been in service in the unit in question for a period of time corresponding to a significant fraction of its expected lifespan (30%, 50%, etc.), for example one year.

[0031] A problem that immediately arose for the person skilled in the art was to have used adsorbent available. In a certain number of cases, it is possible to take samples of adsorbent after several months or years of operation of a unit, or even during a change of load. These samples can be very useful directly if the new unit envisaged differs, for example, from the old one only in the geometry of the adsorber, the new one creating, for example, greater constraints. It is then possible to carry out tests on actually used adsorbents and to check whether they are still compatible with the new constraints.

[0032] It is understood, however, that this case is far from being general and in particular does not correspond in any way to the hypothesis of a new adsorbent (other type, other reference or new supplier) or a new process.

[0033] To compensate for the lack of truly used adsorbent, we have been led to develop so-called "accelerated aging" methods. This involves obtaining in a short time, by an appropriate procedure, a product whose state will be close to that which it would be after long-term operation in an industrial unit. There is no standardized procedure but generally we will play on the frequency of the cycles for cyclic processes (PSA, TSA) and / or on an exacerbation of the physical parameters of the process (pressure, temperature, content of the constituents which may have an influence).

[0034] Some prior arts describe, for example, in great detail a method for obtaining used particles in an accelerated manner. Other aging methods are described later in this document.

[0035] A procedure for selecting the most crush-resistant new and used adsorbents is defined in these prior arts.

[0036] The inventors have however highlighted a parameter which can at least in certain cases completely modify the results of the tests as described in these prior arts and lead to the fact that an adsorbent which would be thus selected could cause significant problems (breakage, dust) during its use in the industrial unit.

[0037] Indeed, it has appeared that the compounds adsorbed during operation could modify the mechanical resistance of an adsorbent and this in a different way depending on the adsorbents. Thus, an adsorbent judged to be the best may in fact turn out to have insufficient mechanical properties for the use that one wishes to make of it, whereas it will be entirely acceptable in other processes involving other constituents.

[0038] This potentially negative effect of the adsorbate (adsorbed compounds) does not only apply to the crushing resistance but to the other mechanical characteristics of the adsorbent (mechanical shocks, temperature or pressure shocks, abrasion, attrition).

[0039] To overcome this risk, a method is therefore proposed which makes it possible to retain only mechanically acceptable adsorbents regardless of the intended use. To do this, we are no longer looking for the most resistant adsorbent available but the adsorbent which will be able to withstand the constraints which will be imposed on it during the operation of the separation unit.

[0040] Before detailing the proposed procedure, it is first necessary to define or recall the meaning of the main specific terms or expressions used below.

[0041] “Qualifying an adsorbent from a mechanical point of view” means verifying by tests that it will retain sufficient physical integrity during normal use in the unit for which it is intended, thus avoiding potential problems related to the presence of debris or dust. Subsequently, in this case, we will simply say of the adsorbent that it retains its physical integrity. By normal use, we mean operation as intended, apart from any incident or operating accident.

[0042] By "adsorber" is meant any type and geometry of envelope used to hold in place an adsorbent mass intended to purify or separate a gas.

[0043] It is recalled that the “normal boiling point (NBP)” is defined as the temperature at which the saturated vapor pressure of a liquid constituent is equal to normal atmospheric pressure, i.e. 1013 hPa. The constituents of the feed gas having relatively high normal boiling points, for example above -20°C, are generally easily adsorbable constituents.

[0044] “Determine the type and value of mechanical stresses to which an adsorbent is subjected during its use" consists of searching for all the events which could lead to the application of a force on any particle and then characterizing this force by its value (for example in Newton) or by the effect which causes it (for example shock at a given speed, vibrations of fixed frequency, amplitude and duration).

[0045] “Determining the mechanical resistance of a particle” consists of determining by adequate tests the limit up to which the particle retains its integrity. According to a variant, this consists of verifying whether the particle retains its integrity - or sufficient integrity - when subjected to a given stress. It should be remembered that the term "rupture" of the particle is generally used to describe any modification of its initial shape that will make it unsuitable for use in the envisaged process, whether these modifications concern its external surface or its volume. In the same way, the terms rupture stress or rupture test can be used for the various phenomena cited.

[0046] By "adequate measurement protocol for determining the mechanical strength of an adsorbent", we mean the tests which make it possible to reproduce the stresses to which an adsorbent is subjected and to evaluate the effects on its physical integrity. These tests are either based on International Standards (ISO, DIN), or developed internally, with the aim of obtaining the value of the mechanical characteristic studied with the desired precision, generally a few percent. This point is detailed in the course of the description of the different aspects of the invention.

[0047] By "batch comprising a representative number of particles", we mean that the result of the test is statistically significant and allows, depending on the case, the determination of an average value and a standard deviation (for example for the measurement of crushing), or a percentage (of dust, of broken particles) then applicable to the population total number of particles. Depending on the mechanical characteristic considered, this could be, for example, 25 particles or batches of 50 grams.

[0048] By "new" adsorbent, we mean an adsorbent as it comes out of its packaging or has only been in operation in the industrial unit for a very short time compared to its expected lifespan, i.e. of the order of one percent for example of this lifespan. We will generally speak of one to a few weeks at most of service.

[0049] By “used” adsorbent, we mean, on the contrary, an adsorbent that has been in service for a significant time, of the order of at least 30% of the expected lifetime, preferably more than 50%. As it is not always possible to obtain samples of adsorbent that have been in service for such a long time within the desired timeframe, those skilled in the art use accelerated aging processes in the laboratory. The term “used” is also used to designate products that have been artificially aged in this way.

[0050] By "saturated" adsorbent is meant an adsorbent that has been in contact with one or more constituents present in the gas to be treated - or an external gas used for regeneration - and that has trapped in its porous network a fraction of these constituents which then constitutes the adsorbate. In conventional processes, the adsorbent is relatively quickly in equilibrium - or near equilibrium - with the vapor phase surrounding it. A new adsorbent in contact with the atmosphere may also contain a significant quantity of water.

[0051] By “regenerated” adsorbent is meant an adsorbent which has undergone treatment in order to eliminate from its porous network the majority of the trapped products described above.

[0052] With these definitions, an adsorbent can therefore be “new regenerated”, “new saturated”, “used regenerated” or “used saturated”.

[0053] A problem then quickly arises for the person skilled in the art in implementing said procedure in the envisaged adsorption separation unit. While there are numerous publications on the thermodynamic properties of adsorbents (capacity, selectivity, kinetics), commercial equipment for measuring these same isotherms, adsorption kinetics, porosities, software for evaluating their respective performances in the various processes, and possibilities for carrying out representative tests of the process in the laboratory or on site, the literature is poor on the mechanical characteristics that the adsorbents must have depending on the applications in which they are used.

[0054] To select adsorbents capable of maintaining their integrity during the operating period planned for an adsorption separation or purification unit (generally from one year to ten years or more depending on the applications), it is necessary to determine all the constraints to which the material will be subjected from its delivery by the supplier until the time when its replacement has been scheduled, generally, as we have just said, several years later. It is then necessary to determine for each of the identified constraints (friction, shock, crushing) the intensity of the constraints to which the material will be subjected and finally determine the corresponding resistance of the material that we wish to use in order to ensure that it is compatible with the use that we want to make of it. Here we use the words "constraint, intensity, resistance" in the general sense. These points will be explained in more detail later.

[0055] The mechanical data usually available for an adsorbent come from the supplier. This will mainly concern the crushing, attrition and abrasion resistances. These values ​​are determined either by the manufacturer's own procedures or according to official norms or standards. It should be understood that this is a quality control for the supplier to ensure that the product complies with its specification. In no case is it a guarantee that the adsorbent will be compatible with the particular application made by its customer, an application that the adsorbent supplier generally does not know in detail.

[0056] It is therefore necessary for the designer of the unit to have a precise method of accepting an adsorbent material from a mechanical point of view when, for example, one wants to change the adsorbent in an already well-known process, when one wants to modify the adsorber or when one develops a new process of separation or purification by adsorption.

[0057] This document describes below the ways of proceeding to obtain the information necessary for the implementation of the proposed validation procedure. This information concerns both the determination of the resistance necessary for a particle to maintain its integrity and the determination of the effective resistance of said particle. This will be validated for the process in question as soon as the second value (effective resistance) is greater than the first (necessary resistance).

[0058] The subject of the invention is a method for mechanically qualifying particles of an adsorbent intended for a process for separation by adsorption of a gas mixture into a most adsorbable fraction and a least adsorbable fraction, the process using at least one adsorber, said method comprising the following steps: a) identifying a set of operating scenarios for the adsorbent, each of which leads to a mechanical stress exerted on the particles, said set of scenarios comprising subjecting the particles to mechanical, thermal or pressure shocks, crushing the particles, setting the particles in motion by a gas flow or by handling, b) determine for each of these scenarios a maximum value of the mechanical stress exerted, depending on the geometry of the adsorber, its mode of rem folding in the adsorbent, of operating parameters of the process, the operating parameters comprising a pressure, a gas flow and a temperature in the adsorber and depending on a variation of these parameters, c) constitute batches of N adsorbent particles, with N > 25, at least one batch of which is made up of new adsorbent and at least one of which is made up of adsorbent having been loaded with the most adsorbable fraction and regenerated repeatedly, so as to obtain an aged adsorbent, d) subjecting the batches formed in step c) to an adsorbent regeneration procedure in order to desorb any impurities, so as to obtain regenerated batches, e) select one of the scenarios identified in step a), f) subjecting the batches regenerated in step d) to a particle rupture test in the scenario selected in step e), so as to determine an average mechanical strength value of the particles of each batch in the selected scenario, g) comparing the average mechanical strength value determined in step f) for a given batch with the maximum value of the mechanical stress determined in step b) for the scenario selected in step e), h) validate the adsorbent for the separation process if the average mechanical strength value determined in step f) is greater than the maximum mechanical stress value for the scenario selected in step e), said method being characterized in that it further comprises the following additional steps: (i) measuring the quantity in the gas mixture of at least one constituent whose boiling point at atmospheric pressure is greater than -20°C, constituents selected from water, hydrocarbons, ethers, alcohols, aldehydes, ketones, j) comparing the quantity measured in step i) with a predetermined threshold value, k) if the quantity measured in step i) is greater than the threshold value, loading the batches regenerated in step d) by constituting it or them, and submitting the batches thus obtained to the testing, comparison and validation steps corresponding to steps f), g), h).

[0059] According to a variant, said process for separating a gas mixture by adsorption is of the TSA, PSA or guard bed type.

[0060] In practice, this validation protocol concerns all gas adsorption treatment units because the adsorbent must always be housed in an envelope in which it will at least undergo the effect of its weight, the inlets / outlets / crossings of the gas flows, and the filling.

[0061] According to a variant, the adsorbent A is part of the group comprising zeolites, with or without binder, exchanged or not, such as, in a non-limiting manner, LSX, X, A, Y, ZSM-5, silicalite, activated carbons of mineral or vegetable origin, optionally doped activated aluminas, silica gels, carbon molecular sieves (CMS), metallo-organic structures, oxides or hydroxides of alkali or alkaline-earth metals, resins, polymer-based materials, or porous structures preferably containing a substance capable of reacting reversibly with the molecules, a substance such as amines, physical solvents, metal complexing agents, metal oxides or hydroxides for example.

[0062] According to a variant, the adsorbent A tested corresponds exactly to the reference of the adsorbent which will be used in the separation process.

[0063] Indeed, the same category of adsorbent, for example a zeolite X, will have different characteristics depending on the suppliers due to the raw materials and manufacturing protocol specific to each of them. In particular, shaping using additives such as mineral binders or binders of the resin or polymer type to which elements intended to modify the porosity or other physical characteristics may be added, the various post-treatments possibly applied to reduce the reactivity, to modify the surface state can modify the mechanical characteristics of the particle.

[0064] According to a variant, the constituent X having a normal boiling point (NBP) greater than -20°C is part of a group comprising in particular water, hydrocarbons, ethers, alcohols, aldehydes, ketones.

[0065] These are relatively frequent constituents in the gases to be treated, often in the form of impurities which are retained mainly in the first beds of an adsorbent mass, upstream of adsorbents more specific to the separation process implemented. This may involve stopping atmospheric humidity for example before an O2 / N2 separation, or stopping traces of methanol before decarbonation. Some of these constituents may be found in extremely small quantities in the gas mixture and therefore have no significant effect on the mechanical strength of the adsorbent particles. It is appropriate to set a threshold from which the various X constituents mentioned could become problematic. Setting a limit content or partial pressure is not a practical approach because this would be closely linked to the nature of the impurity and that of the adsorbent.It is proposed to favor the following approach to decide whether tests with "polluted" adsorbents are necessary before definitively validating an adsorbent. From the quantity of component X in the gas mixture and the pressure and temperature characteristics of the process, its maximum adsorbed quantity is determined as a percentage by weight of adsorbent A and additional tests are only carried out when the adsorbed quantity X is greater than 1% by weight.

[0066] The adsorbed quantity is determined via the isotherm of the component X on the adsorbent A, a priori well known if we dimension said unit. It is also possible to make a measurement in the laboratory to obtain this value.

[0067] It should be noted that even if a first bed is used to stop these constituents, a bed whose good mechanical performance in the presence of said X constituents can be verified, often the final fraction of these impurities, for example the last percent of the incoming quantity, is stopped at the very beginning of the next bed. This is for example the case of air drying-decarbonation on a double bed comprising activated, doped or standard alumina followed by a zeolite. It is frequently accepted that activated alumina adsorbs the vast majority of the water, the last percent or percents of the water being adsorbed on the first centimeters of zeolite, in particular after several months of operation. It is therefore also advisable to test the zeolite containing a little moisture from a mechanical point of view to ensure that, in this state, the latter does not lose too much of its properties.Using the corresponding isotherm, it can be verified that this zeolite may contain a few percent by weight of water even with contents of the order of ppm. It can be noted that in the case of a radial bed, it is this part of the zeolite bed which is in direct contact with the intermediate grid of the adsorber and which will therefore undergo particularly high stresses. To avoid this problem, it may be interesting, in order to validate an adsorbent with too limited mechanical resistance in the presence of an impurity, to take a margin during the design such that the impurity in question never reaches the bed located downstream containing the said adsorbent. For example, a design will be carried out for a complete shutdown in the case of the most unfavorable operation with an adsorbent whose adsorption capacity would be reduced by 20% due to aging.According to a variant, the content of the constituent X in the adsorbent is advantageously determined spatially and temporally along the adsorber by means of adsorption process simulation software adapted to the separation process.

[0068] Indeed, the nature of the adsorbed constituents and their quantity will depend not only on the composition of the gas to be treated but also on the process envisaged and the position of the adsorbent in the adsorber. For example, particles located at the inlet will be in contact with all the constituents of the feed gas throughout the production phase while conversely particles located at the outlet will only see purified gas. As the compositions change over time depending on the stages of the cycle, it will generally be necessary, for greater precision, to use the results of adsorption process simulation software to determine the adsorbate as a function of time and the position of the particles in the adsorber.

[0069] The above-mentioned dynamic simulation software for cyclic adsorption processes is either a “proprietary” simulator developed internally by the designer of the unit, or one of the commercial software. It is based on a division in the space of the adsorber and in the time of the cycle and performs successive sequential calculations until a stable state is found verifying material and thermal balances as well as all the imposed constraints. For a given application, in addition to the details of the cycle, it contains the physical properties of the gas flows (viscosity, heat capacity, conductivity, etc.), the detailed geometry of the adsorbers, the characteristics of the adsorbents (dimensions, isotherms, kinetics). The input data for performing a simulation depend on the process.We generally find all the data concerning the feed gas (composition, flow rate, pressure, temperature), the different levels of internal pressure in the unit, the nominal duration of the stages, with generally a free parameter allowing the simulation to converge on the main constraint (for example modification of the phase time or the flow rate of feed gas to respect the purity required for production). The detailed description of such software exists in the literature and will not be repeated here.

[0070] According to a variant, the particles of used adsorbent used in the tests in order to determine their mechanical resistance were taken after an operating time of several weeks and preferably several months on an industrial unit (or on a small test unit) characteristic of the separation process.

[0071] By a unit characteristic of the separation process that one wishes to use, one means a unit whose operating parameters are sufficiently close to said process so that the adsorbent experiences over time the same type of effects linked in particular to the composition, the temperature, the number of cycles. It is appropriate that the contaminants that could play a role in possible degradation (such as humidity for example) are present, that the temperature levels are within plus or minus 10 or 20°C for example and that the operating time corresponds to at least 30% of the expected lifetime for the separation unit envisaged.

[0072] According to a variant, the used adsorbent particles used in the tests were aged in an accelerated manner.

[0073] The accelerated aging procedure consists of subjecting the adsorbent to successive adsorption / regeneration cycles with a frequency at least ten times higher than the actual cycle or of subjecting it to temperatures and / or pressures going beyond those of industrial operation and / or impurity contents X higher than that corresponding to the feed gas of the unit. Hybrid procedures, i.e. accelerated cycles with temperature and / or pressure conditions more restrictive than those of the process, can also be used. Such procedures as well as the associated equipment are described in detail below.

[0074] As indicated, the mechanical constraints for which the resistance R of the adsorbent is measured are part of the group comprising mechanical shocks, attrition, abrasion, thermal shocks, pressure shocks or crushing. These mechanical constraints correspond, depending on the intended application, to the filling of the adsorber, to its possible transport, to one - or more - stages of the adsorption process, in particular to the stages of adsorption, desorption, depressurization, pressurization, heating, cooling, and possibly to the emptying of the adsorber.

[0075] The protocols for measuring resistance R are either protocols corresponding to international standards (DIN, ASTM), or specific protocols developed internally more adapted to the specific case of adsorbents. In particular, within the framework of these protocols, the constraints applied to the particles, covering in a sufficiently broad way the range of those which they may have to undergo, must give measurable and differentiating results. Some of the international standards, due to their general nature, are not sufficiently relevant for adsorbents with insignificant results (for example: almost non-measurable quantity of dust).Therefore, just as adsorbent suppliers often have their own measurement protocol associated with their specifications, designers of adsorption gas separation units have generally developed tests adapted to the adsorbents used, that is to say, they allow, as indicated, a quantification of the mechanical characteristics and a differentiation between adsorbents. For example, attrition will be measured using a vibrating sieve whose amplitude, frequency and duration of vibrations will have been fixed by experience.

[0076] Some of the methods that can be used are described in the example chosen to illustrate this validation procedure. This will be a TSA comprising radial adsorbers.

[0077] The resistance values ​​T necessary to validate an adsorbent are obtained either by experience taken from similar units in which adsorbents already tested according to the same protocols and whose limits are known are used, or by stress calculations based on the geometry of the adsorber and the process implemented, or by a mixture of these two methods.

[0078] This is one of the most important points of the procedure. Indeed, retaining, as is sometimes recommended, the most resistant one from a group of potential adsorbents for a given process can present numerous disadvantages. We have seen that if we are required to test several mechanical properties, such as for example resistance to crushing and abrasion, we can find ourselves with a dilemma and with two adsorbents, each being better in a category.

[0079] On the other hand, we can also find ourselves in the case where the best of the adsorbents does not have sufficient resistance or conversely that the adsorbent coming in second or third position has sufficient mechanical resistance to be used. in the industrial unit and superior separation properties than the one that will be retained as the best. In the first case, the industrial unit will experience serious incidents, in the second its performance will be below what it could be.

[0080] It is therefore appropriate to determine the mechanical resistance values ​​to be achieved in order to avoid the pitfalls mentioned above. When it comes to very conventional processes implemented for decades, such as air purification before a cryogenic unit, the designer has a very large database comprising various adsorbents, various variants of the process and different types of adsorbers. From all these experiences, it is generally possible to derive the minimum resistance values ​​for which no problems were reported during operation and where, during emptying, neither debris nor dust in abnormal quantities were found. Particularly interesting cases correspond to the identification of areas with higher than normal quantities of broken particles or dust. It is then possible to have a very good idea of ​​the limit values ​​in this operating case.This can help validate or calibrate theoretical calculations when conditions are far removed from what is known (significantly different process, unusual adsorber shape). We will return to this point using the example already mentioned of a TSA with radial adsorbers.

[0081] It will also be noted that if the presence of an adsorbate often leads to a reduction in mechanical resistance with respect to at least one constraint (crushing, abrasion), the opposite, i.e. the case where the adsorption of a constituent would increase the mechanical resistance of an adsorbent, seems to be very rare or even improbable. Thus, by limiting oneself to testing adsorbents in the saturated state, when these are actually in contact with a constituent X having a normal boiling point (NBP) greater than -20°C, one is a priori on the safe side with respect to their use in the separation process, nevertheless one would then deprive oneself of useful information that step b) of the proposed protocol makes it possible to obtain.

[0082] When it is found that the presence of constituent X actually has an effect on the mechanical strength of adsorbent A, it is interesting to test this same adsorbent, new and used, brought into contact with different partial pressures of constituent X in order to determine the effect of the quantity of adsorbate on the mechanical strength of the adsorbent. This test is particularly simple to carry out in the case where constituent X is water because there are well-known methods for creating different relative humidities using various salts in aqueous solution.

[0083] Depending on the case, a small quantity of adsorbate is sufficient to significantly modify the mechanical resistance of the adsorbent or, on the contrary, the reduction is relatively proportional to the quantity adsorbed and the adsorption of traces has little effect. This information may be necessary to determine the most critical zone for an adsorbent in contact with a constituent X knowing that the mechanical stresses and the impurity contents will vary along the adsorber.

[0084] Given the diversity of adsorption separation processes, the adsorbents that can be used in said processes, and the shapes and dimensions of the adsorbers, it is difficult to predict with certainty what the most restrictive constraint is in the choice of an adsorbent. Adsorbent particles may, for example, be particularly resistant to crushing due to a robust internal structure but have a tendency to wear on the surface by friction or to break following an impact (brittle particle).

[0085] Said mechanical constraints correspond, depending on the application envisaged, to different events or moments such as for example the filling of the adsorber, its possible transport, one - or more - steps of the adsorption process, in particular the steps of adsorption, desorption, depressurization, pressurization, heating, cooling, and possibly emptying of the adsorber.

[0086] The person skilled in the art will determine the set of tests to be carried out to ensure normal industrial operation depending on the characteristics of the unit. For example, we will generally not be interested in thermal shocks in the case of PSA at ambient temperature, in pressure shocks in the case of TSA, in mechanical shocks in the case of low height adsorber (1 to 2 m) and gravity filling.

[0087] It is important to have aged adsorbents because they generally have lower mechanical strength than new ones. Since it is not easy to obtain samples that have spent several months in an operating unit, procedures have been developed to "artificially" age these materials. The methods used, although based on the same principles (accelerated cycles, increased temperature, pressure, composition effects), are specific to each adsorption separation unit designer.

[0088] We have seen that one of the known ways to obtain particles of used adsorbent was to subject small samples of adsorbent to accelerated aging cycles. This involves either carrying out generally simplified cycles with a frequency much higher than the cycle according to the industrial process envisaged, or maintaining these samples for a certain time in conditions much more restrictive than those of real operation. For example, a TSA type cycle will take place in one minute (adsorption and regeneration) instead of lasting four hours industrially. Thus, in 48 hours, more than a year of real operation can be simulated. In practice, this type of test is generally carried out in the laboratory using benches entirely automated and may include, where possible, several channels in parallel. It is thus possible to process different samples simultaneously, either to obtain information on several different products, or to determine the variability between several samples of the same product. For a given type of test, the person skilled in the art will first check the impact of the choice of parameters retained in the test protocol. For example, the impact of the cycle duration on the physical property(ies) to be checked will be determined, namely the possible difference between a sample having carried out 100 one-minute cycles or 100 five-minute cycles. This type of information is then used to modify the test protocol and possibly to add corrective factors. These tests are also used to determine the degradation kinetics by testing particles having carried out different numbers of cycles.This degradation can be linear with the number of cycles, accelerate over time or conversely tend towards an asymptote. There are no general rules and it is usually appropriate to obtain several experimental points under different conditions in order to determine the trend of the curve. From this type of study, it is possible to determine not only whether an adsorbent is actually usable in a given process but also an order of magnitude of its lifetime.

[0089] [Fig.l] represents a test bench; [Fig.2] represents an aging bench of an adsorbent; [Fig.3] represents an adsorber.

[0090] [Fig.l] shows the operating principle of such a bench. It should be noted that this is only an approach to real operating conditions and that it is mainly a question of comparing the behavior of different adsorbents, some of which, if possible, will already have been successfully implemented in neighboring processes. This bench is particularly useful in this respect for validating a new adsorbent.

[0091] In order to limit the workload as much as possible, bench 1 is entirely managed by a programmable logic controller 2 and all the information is stored, the main parameters being permanently displayed on screen or printer 3. The bench comprises six cells or adsorbers 11, 12,... 16 which can contain adsorbent and operate essentially in parallel. These adsorbers are in a thermally insulated enclosure 6. The adsorbers in operation are successively supplied with water-saturated nitrogen at a pressure P1 and a temperature T1 then with dry and hot nitrogen at a pressure P2 and a temperature T2. The water saturation takes place in a thermostatically controlled cell 4 while the regeneration gas passes through an electric heater 7. The pipes are thermally insulated with a thermal insulator 5.The feed and regeneration flow rates are respectively DI = N*dl and D2 = N*d2 with N the number of adsorbers in operation (here N less than or equal to 6) and dl and d2 the unit flow rates of each small adsorber. The flow rates DI and D2 are . regulated by the PLC via valves VI and V2. In the diagram of [Fig.l], the flow rates DI and D2 are constant and either directed towards the adsorbers 11 to 16, or towards a vent via the 3-way valves V3 and V4. This makes it possible to reduce the duration of the transient phases in the case of short cycles. Each adsorber has on its circuit an orifice caliber 21, 22, ... 26 allowing a distribution of the total flow rate into identical flow rates dl, d2 towards the different adsorbers. Each adsorber also has upstream isolation valves 31 to 36 and downstream 41 to 46 automatically controlled by the PLC as well as an unclipping system allowing the adsorber to be removed. The various temperature, pressure, differential pressure sensors are not shown on the diagram.

[0092] Generally speaking, a test is carried out according to the following procedure. The six adsorbers are filled with the same mass of particles, generally from about ten grams to about one hundred grams. The nitrogen flow rates are determined to respectively saturate and regenerate the adsorbent mass in the given times T1 and T2, for example 60 seconds each. Depending on what is being sought, it is possible to choose to test simultaneously from one to six different adsorbents. It is preferable to limit one to three to have redundancy in the results and in this case, the six adsorbers will be filled, in groups of two, with each of the three adsorbents selected, which allows an initial analysis of the reproducibility of the results obtained. It will then be possible to carry out two thousand cycles in about three days of operation and compare the mechanical strength of these different adsorbents to possibly eliminate some of them.If we want to validate more definitively an adsorbent that seems interesting, we can proceed differently. We generally put five samples of the new adsorbent accompanied by a control sample of an adsorbent as a check of the correct functioning of the bench. The automaton is then programmed to isolate one of the adsorbers after, for example, 100 cycles, a second after 500 cycles, then 1000, 1500, the last two - including the control - after 2000 or 2500 cycles. We can thus determine the kinetics of degradation of the adsorbent with respect to a particular test (shock, attrition, crushing but also adsorption capacity) as a function of a number of cycles. We can observe linear, exponential degradations or on the contrary tending towards an asymptote and this to more or less significant degrees.

[0093] While aging by the effect of humidity is the simplest to achieve, it is possible to use other components that can be injected in various ways (membrane, metering pump, saturator). It will of course be necessary to adapt the safety devices to the products used (flaring, atmosphere analyzers with alarm, confinement).

[0094] [Fig.l] shows an aging bench installed in a laboratory. In order to study the impact of water and temperature on an adsorbent, the gas supply may only have a secondary effect and it is then common to use nitrogen, dry air or a simple mixture of gases. When we want to test a gas close to an industrial gas containing many compounds, the system for supplying the bench with reconstituted gas similar to the real gas can have many problems. This "reconstituted" supply is obtained by mixing pure gases - or already partially combined - which are delivered in cylinders. Depending on the number of constituents retained, we may have to use a rack of five or six cylinders - or even more - of different composition to be used in parallel. The cost of such a supply quickly becomes high. This also requires complex regulation systems, precise analyses and safety procedures developed according to the compounds present (health risks, explosion).A second disadvantage is that in most applications, the feed gas thus "reconstituted" does not contain all the impurities found in the actual gas. However, we know that some problems come from impurities present only in trace form. It may therefore be necessary to carry out on-site tests using actual gases. This may be the case in particular for biogas treatment units, especially if the latter comes from buried landfills: traces of several dozen compounds resulting from the fermentation or decomposition of waste are generally found.

[0095] For this reason, small units, generally transportable, have been developed in parallel with laboratory units, which can be installed on the industrial site itself. These test benches can range from the simplest, namely tubes filled with potential adsorbents continuously swept by the gas, to the most complex, namely small PSA or TSA units. These benches have all the necessary equipment to operate autonomously without special supervision. Depending on the case, we will therefore find valves, various sensors, analyzers, control-command system, data recording, storage and processing system). The gas to be treated is taken directly from one or more pipes of the industrial units located in the vicinity of the test unit. In this way, if deemed useful, the composition of the feed gas can be varied by modifying the ratios of the mixtures.It may also be interesting to add to the mixtures obtained from the gases from the industrial units one or more contaminants whose behavior we want to evaluate and which would be absent or in too low a content on the site chosen to carry out the tests. Since the flow rates sampled are very low compared to the flow rates treated on site, the connections are generally made at the various existing sockets on the site's pipes (analysis sockets, various sensors, purges). Similarly, the gases from the test unit are generally returned to a so-called "fuel-gas" network, a flare network or any other suitable pipework. These test units, which are generally designed to be mobile, are housed, for example, in containers. The equipment. Electrical or electronic equipment placed near the units are generally chosen to be used safely in an industrial environment. Other equipment can be moved to a safe area if necessary. [Fig.2] shows schematically such an installation intended to obtain samples of used adsorbents, i.e., those swept for several months by the gas whose effects on these materials likely to be used in place of current adsorbents are to be known. In the industrial process, the humidity contained in the feed gas is stopped in a first layer of adsorbent and it is the adsorbents coming after this guard layer which are the subject of the test.In order not to distort the results by the presence of humidity, the small flow sampled 10, at the level of an analysis socket 8 located on the main pipe 56 is dried through a desiccator 17 before being sent in parallel to the series of tubes 18, 19, 20 containing the new adsorbents 51, 52, 53. The desiccator 17 will be preferably filled with zeolite 3A (in English: deeply exchanged), that is to say containing in practice only crystals of 3A in order not to adsorb any other contaminant and thus not to modify the composition of the gas to be treated. A bottle 54 containing a compound suspected of playing a role in the premature aging of the adsorbents is connected to the tubing supplying the tube 20 making it possible to artificially increase the content of this constituent.It will then be possible, by modifying the injected flow rate of this additional impurity, to highlight the impact of the partial pressure of this constituent on the characteristics of the adsorbent. The gas having passed through the adsorbers can be discharged for example into a low-pressure residual gas pipe 55. The various solenoid valves, calibrated distribution orifices, sensors are not shown in this figure. In [Fig.2], the additional compound is injected from the pressurized gas cylinder 54. Depending on its physical characteristics, this compound can be pure or mixed with a constituent of the main gas. Similarly, it could be in liquid form and injected by means of a metering pump or any other suitable system (membrane, saturator).

[0096] The second type of test, called "accelerated aging", consists of placing the adsorbent in a situation that is significantly more unfavorable than the industrial one. It is then more difficult to extrapolate a lifetime under normal operating conditions, but this makes it possible to compare various adsorbents, including, if possible, at least one of them favorably referenced for said application. In this second type of test, the adsorbent is generally placed in an autoclave in order to be in contact with the vapor pressure of the selected constituent likely to degrade the adsorbent at least in the long term. The temperature and duration of this manipulation can then be varied.

[0097] The advantage of having actually used adsorbent taken from the an industrial unit. Such a sample makes it possible to determine the effect of a known operating time on resistance to a stress, for example, crushing resistance. Let us suppose that a 25% decrease in this resistance is noted compared to the same new product. This information will make it possible to retain the conditions of the accelerated aging protocol which, for this same adsorbent at least, lead to a decrease of the same type, namely here of the order of 25%. If the accelerated aging consists of a residence time t in an autoclave brought to a temperature T, we will retain, for example, two pairs of parameters, tl and Tl on the one hand and t2 and T2 on the other hand - with tl>t2 and T1 <T2 - qui conduisent à des résultats correspondant à une perte proche des 25% mesurés.

[0098] This protocol will then be extended to other adsorbents, which will allow us to get a good idea of ​​their residual resistance under these conditions, or at least differentiate the behavior of the various adsorbents.

[0099] To now illustrate the invention in detail, we take the example of a TSA process for drying and decarbonating a gas which follows a two-hour cycle, approximately twice as short as usual cycles, comprising two adsorbers, one being in regeneration when the other is in adsorption. These are radial adsorbers whose dimensions (height, diameter) are large but remain within the standards already achieved. The adsorption pressure is relatively high (20 bara) but also remains within a usual range. Due to the shortening of the cycle and the operating pressure, the pressurization and depressurization speeds will be faster than usual. The regeneration temperature is a little hotter than usual, of the order of 175°C, in order to minimize the necessary regeneration flow rate. On the adsorbent side, it is a double bed, activated alumina followed by a zeolite.Activated alumina is an adsorbent from a new supplier, economically interesting but not used in this application, zeolite is a product already used conventionally for similar applications. This example will not exhaustively cover all the points to consider in this type of approach but is intended to illustrate the steps to follow to avoid potential incidents in this unit.

[0100] An adsorber 60 of the type used is shown in section in [Fig. 3]. The straight part of the adsorber 60 is shown, the axis of symmetry of which corresponds to the broken line AA. The shell 61, the upper bottom 62, the lower bottom 63 on which the feed gas inlet and waste gas outlet pipe 64 are fixed, as well as the eccentric production gas outlet and regeneration gas inlet pipe 65, are shown. A method has been chosen requiring two successive layers of adsorbent 66 and 67, conventionally arranged in two juxtaposed annular beds, separated by an intermediate grid. Thus 66 is the reference of the first bed and 67 that of the second, beds held in place by means of the external grid 68, intermediate 69 and internal 70. This type of adsorber is suitable for many applications, particularly when treating large flow rates (several tens of thousands of m3 / h), and / or when we want to minimize pressure losses or, conversely, when we want to circulate at high speed without setting the adsorbent in motion.

[0101] The diameters of the shell 61, of the external grid 68, intermediate 69 and internal 70 as well as the height of these elements are very variable depending on the applications. Generally, the shell can thus have a diameter of 1 to 7m, more generally of 2 to 6m and its height between 3 and 25m. Here, an adsorber 60 of elongated shape of 4m in diameter and about twenty meters high is used for the part containing the adsorbent mass. The inlet pipe 64 passes vertically through the lower bottom 63 of the adsorber 60 along its axis of symmetry AA and is extended inside by a tube 71 up to approximately the upper level of the grids, at which level it is closed by a solid disc 76.The grids are fixed at their lower end to a ring 72 which is essentially flat, the large diameter of which is approximately equal to that of the external grid, this disc being on the one hand welded to the inlet pipe 64 at the level of its small diameter and on the other hand resting on a cylindrical ferrule 73 itself resting on the lower bottom 63 of the adsorber 60. This solid ring also constitutes the lower sealing means of the annular volumes housing the adsorbent masses.

[0102] Facing the internal grid 70 and over its entire height, the tube 71 extending the inlet pipe 64 - of diameter smaller than the diameter of the grid - is perforated. This perforation 74 allows a first distribution of the fluid both axially and radially in the inter-wall 75. The solid disc 76 seals the upper end of the pipe 71.

[0103] To maintain its freedom of axial movement, the external grid 68 is centered at its upper part by a guide system fixed to the adsorber 60, preferably by a cylindrical ferrule 77 welded to the upper bottom of the adsorber 60 and of a diameter slightly different, here greater, than that of the grid. A solid disc 78 of diameter equal to that of the internal grid closes the upper end of the internal grid.

[0104] The sealing in the upper part of the grids is achieved by a flexible wall 79 which is essentially horizontal and kept pressed against the particulate material by a pressure effect. In practice, this is a flexible metal membrane 80 which is essentially circular and fixed to the upper non-perforated part of the external grid and kept in contact with the adsorbent by means of a mass of metal balls 82 such that the membrane remains in contact with the upper free surface of the adsorbents during all stages of the cycle. Orifices 81 arranged in the upper part of the grid external allow the pressures to be balanced between the external distribution volume 84 and the dead volume at the adsorber head.

[0105] As described above, the inlet 64 is made by a central tube entering the adsorber 60 through the lower bottom. Taking into account the general arrangement adopted, the outlet tube 5 is also fixed, in an off-axis manner, to the lower bottom 63 of the adsorber 60. It is located inside the space between the support cylinder 73 of the ring 72 on which the grids are fixed and the central tube 64. The support cylinder 73 is obviously permeable to fluids and the perforation rate 83 is determined in order to ensure satisfactory distribution of said fluids.

[0106] The problem here therefore consists of mechanically validating a new activated alumina which must be used essentially as a first guard layer to stop the vast majority of water (>99%) contained in the gas to be treated. The second adsorbent is a zeolite already referenced for stopping CO2 but under slightly different conditions. We will therefore have to take into account the characteristics of the adsorber 60 used as well as the details of the process to determine whether the two adsorbents selected, a priori on the separation aspect alone, are compatible with the constraints they will have to undergo.

[0107] We will therefore, by reviewing the operating scenarios that can create constraints on the particles and cause their rupture in the sense defined above, successively take into account the risks linked to mechanical shocks, abrasion / attrition and crushing resistance which will be studied in detail because a priori this will be the most important constraint given the specificities of the unit. Although thermal and pressure shocks do not appear to be critical given the experience acquired previously, these tests will also be discussed below for information purposes.

[0108] For each of the tests envisaged, it is appropriate to define the state of the adsorbent to be tested and the stress that it will have to withstand.

[0109] It is recalled that the state of an adsorbent - as this term is used here - corresponds to two very different characteristics: on the one hand its duration of use in the process (one day, one week, one year, etc.), that is to say in practice "new or used", on the other hand what it contains, that is to say the nature and quantity of the constituents which are adsorbed (the adsorbate), that is to say in practice, according to the vocabulary adopted, "regenerated or saturated".

[0110] As already indicated, the nature of the adsorbed constituents and their quantity will depend not only on the composition of the gas to be treated, but also on the process envisaged and the position of the adsorbent in the adsorber. For example, particles located at the inlet will be in contact with all the constituents of the feed gas during the entire production phase and most of the regeneration stages, whereas conversely, particles located at the outlet will only see purified gas.

[0111] Since the compositions change over time depending on the stages of the cycle, it will often be very interesting, as already taught, to use the results of adsorption process simulation software to determine the adsorbate as a function of time and the position of the particles in the adsorber.

[0112] The next point is to have available an adsorbent in the saturated state such that this state could be determined using the simulation, whether this adsorbent is new or used.

[0113] It should first be noted that for the vast majority of industrial adsorbents the presence in the pores or active sites of weakly adsorbed gas (Ar, CO, H2, He, N2, O2) has no impact on their mechanical strength. The same is true for constituents which may be significantly more adsorbed, such as CH4, or even CO2 for example, but which remain in the majority of applications far from a possible state close to liquid. This is all the more true since most applications take place at moderate pressure, generally below 35 bar absolute and rarely above 60 bar absolute. For applications at higher pressures, in particular 100 bar absolute and above or families of adsorbents with a flexible framework, these conclusions should be reviewed.

[0114] We are therefore limited here to taking into account the constituents which can potentially have a negative effect on the mechanical resistance, namely essentially constituents having low vapor pressure and / or adsorbing very strongly due to their interaction with the adsorbent (very polar bodies for example). It is difficult to give a general rule on this point, this phenomenon depending not only on the adsorbent / adsorbate pair, but also on the manufacturing characteristics of said adsorbent: nature and quantity of the binder, process for shaping the particle, size of the pores and the quantity of adsorbate in the particle.It is possible that the presence of capillary condensation in the pores of smaller diameter plays a role in the modification of the mechanical properties of the adsorbent, but other phenomena can play important roles (interaction of the adsorbate with the structure of the binder, with the framework of the active part, increase in internal pressure, etc.). In the absence of precise rules, it is appropriate to take into account a possible degradation of the mechanical resistance characteristics, as soon as there are constituents in the feed gas with relatively high normal boiling temperatures, for example above -20°C, more particularly above 0°C.

[0115] In the example chosen, only humidity (possibly associated with temperature) must be taken into account as a factor in both aging and mechanical weakening of the adsorbents.

[0116] Depending on its position in the adsorber, the alumina will contain water at its turation or only in traces. A small fraction of the zeolite bed will see the residual water. These conditions are classic and common to many units.

[0117] Obtaining "used" adsorbent is carried out using the bench described previously and corresponding to [Fig.2]. Only activated alumina is aged in this way because it is a new adsorbent for this application. Zeolite has already been the subject of numerous studies of this type.

[0118] Independently of the mechanical resistance, it appears from these tests that the different activated aluminas also lose their adsorption capacity. This degradation is a function of the quantity of water, the temperature, the duration of operation and the adsorbent itself. It is customary to attribute this loss to a chemical modification of the structure of the alumina, a fraction of which would become almost inert. This is generally referred to as hydrothermal degradation. Even if this is a general phenomenon, its intensity depends strongly on the activated alumina. Losses of adsorption capacity can thus be observed, under the same conditions, ranging from 10 to 35%. This shows that the stability of this adsorbent depends a priori on the raw materials used and the manufacturing process. This information encourages systematic testing of any new reference of activated alumina.The activated alumina here is average with a capacity loss of around 20% which was taken into account when sizing the unit.

[0119] It is recalled that the principle of the invention consists in measuring the mechanical resistance of an adsorbent in the presence of its adsorbate, the latter being likely to significantly modify the results obtained with the same regenerated adsorbent. In the case of humidity, the procedure for obtaining samples with various water contents is particularly simple. Adsorbents in equilibrium with different relative humidities are obtained by using "salt boxes". Indeed, by using thermostatically controlled sealed boxes (generally ranging from 5 to 60°C) containing various saturated saline solutions, it is possible to obtain at equilibrium a gas phase whose relative humidity RH can vary from 10 to 90%. The adsorbent samples are placed inside these boxes. In this way, around ten different salts can be used, including, for example, lithium chlorides (HR# 10-15%), magnesium chlorides (HR# 33%), sodium chlorides (HR# 75%), potassium chlorides (HR# 85%).This allows us to obtain a degradation curve with respect to the water contained in the adsorbent.

[0120] Having “new” and “used” adsorbents which can easily be obtained in the “regenerated” state by sweeping with a hot and dry fluid (the regeneration temperatures and necessary sweeping times are known to those skilled in the art but if necessary, it is possible to verify that the regeneration is complete by analysis or weighing) and in the “saturated” state via a simple method, it is possible to evaluate their mechanical resistance.

[0121] For each of the constraints that can pose problems (mechanical shocks, attrition and abrasion, thermal shocks, pressure shocks, crushing), it is therefore necessary to agree on a test and the acceptable limit value of the result obtained. It is also necessary to determine, as just explained, the state - or states - of the adsorbent to be tested (new regenerated, new saturated, used regenerated, used saturated). This must be done constraint by constraint.

[0122] The stress generally simplest to study is that corresponding to a mechanical shock. A "mechanical shock" is a sudden contact between an adsorbent particle and a rigid surface. Very generally, it is the particle moving at a certain speed which strikes the stationary surface. Referring to [Fig.3] concerning the example, the maximum stress will correspond to the start of filling the adsorber when the first particles strike the metal bottom 72 which serves as a seal or strike the grid 68, 69, 70 intended to hold them in place laterally. The shock speed will depend on the filling method in addition to the dimensions of the adsorber. In order to achieve dense filling, the final fall speed may correspond to a free fall of the height of the adsorbent bed possibly increased by the effect of injecting a gas into the filling system.While it is theoretically possible to reproduce such a fall experimentally, it is more practical to calculate the speed at the time of impact and to project particles at the appropriate speed onto a metal part or a piece of grid. For this first test, we understand that the state of the adsorbent will a priori be the "new regenerated" state as taken from the supplier's packaging or "new saturated", if at least part of the adsorbent is in equilibrium with the humidity of the atmospheric air, depending on the storage conditions and local atmospheric conditions. The acceptance criterion for an adsorbent will then be a maximum weight percentage of broken particles (for example 1%) collected by sieving after an impact at a speed for example 25% higher than the calculated speed, which corresponds to a margin of more than 50% on the impact energy. Samples of for example 50g can be taken, which already corresponds to a large number of particles.For the sole purpose of comparison between adsorbents, it may be agreed to add a test under fixed conditions, independent of the unit in question. This allows for a library of comparable data but also to possibly detect, for the same adsorbent, variations over time. This remark on the interest of having for comparison on identical bases, a standardized test, independent of the real conditions of future operation, applies to all the characteristics that may be required to be measured.

[0123] In the case which concerns us, we will analyze the result of impacts of activated alumina particles coming into contact with a wall at speeds of the order of 10 to 20m / s For example, we will test a sample of adsorbents taken from the free surface of the packaging containing the activated alumina, which may therefore contain a little moisture ("saturated particles"), and a sample of adsorbents taken several centimeters below the surface and therefore protected from potential atmospheric ingress (regenerated particles). We can also deliberately leave a sample of activated alumina in the open air for a certain time, for example one hour, if such contact can occur during filling. We will thus be able to compare the resistance of new "saturated" samples and new "regenerated" samples by the supplier.

[0124] The new alumina passes this test without any particular problem with a percentage of debris significantly lower than the standards adopted. Some silica gels, among industrial adsorbents, can be relatively brittle for example, but standard aluminas, zeolites and activated carbons are generally relatively resistant to shocks of this type. Nevertheless, the search for high kinetics can lead to the use of either a minimum of binder or pore-forming elements intended by post-treatment to increase the pore volume and it is advisable to be vigilant in particular for new generations of adsorbent with increased performance.

[0125] The adsorbent particle must also be able to withstand a certain form of wear caused by friction and referred to as abrasion and attrition. This refers respectively to the propensity of the adsorbent to wear on the surface, producing (fine) dust or small chips when subjected to friction. There are many tests, standardized or not, to measure these phenomena. Usually, suppliers carry out their own test on "new regenerated" adsorbent and guarantee a maximum weight loss value at the end of the test. It should be noted that apart from adsorption processes in which the adsorbent is deliberately set in motion, the vast majority of units are of the fixed bed type. The person skilled in the art who designs adsorbers takes care to avoid speed zones which would set the particles in motion.However, despite these precautions, abrasion or even attrition may occur during transport or operation, locally, on the free surface, on the wall or in small cavities that may form due to the silo effect. In the case of a new adsorbent, a new supplier - as is the case here - or a new process, it is advisable to verify by internal tests that the material under consideration does not deviate from the materials already referenced. It is advisable to test the new and used product, both in the saturated and regenerated state.

[0126] A used or saturated adsorbent (by water, alcohol, heavy hydrocarbons) can be "weakened" at least on the surface and behave in a significantly different way depending on its condition. As already indicated, tests will preferably be used under fixed conditions in order to be able to compare on the same basis the large number of adsorbents that may be used. A simple test consists ofsubject a sample of particles, for example 50 g, to vibrations of a given frequency and amplitude for a fixed duration, for example 30 minutes. At the end of the test, the dust generated by the vibrations is sieved and weighed. The operating parameters of the vibrator are chosen so that for the majority of adsorbents, a percentage of fines of the order of 0.1 to 1% by weight is obtained, that is to say both measurable and of the order of magnitude of what can be obtained locally in an industrial unit. This is a comparative type test intended to ensure that an adsorbent that one wishes to implement is, by experience, in the correct range of values. It would indeed be very difficult or even impossible to determine the precise conditions of wear (frequency, amplitude, contact surface) of a particle, these conditions being able to vary from one location to another in the adsorber.It should be noted that for the same product, from the same supplier, there may be surface post-treatments that make it possible to obtain adsorbents that are more resistant to attrition without any significant loss of other physical properties. This is particularly the case for activated alumina. As in our example, the operation is a little more critical with regard to these phenomena - due to locally high radial speeds - than in the general case, it may be in our interest for safety reasons to choose the grade with increased resistance to attrition despite a slightly higher cost.

[0127] It is noted that new activated alumina, new or used, saturated with water tends to produce more dust than when it is in the regenerated state and that it produces more than the reference product while remaining within the acceptability range. It should be noted that although the principle of the measurement is simple, its implementation requires particular care (initial mass previously blown to eliminate residual fines, taking into account humidity in the calculation of masses).

[0128] Depending on the applications, the adsorbent may be subjected to thermal shocks: this means both processes using cold temperatures (for example in the case of cryogenic purification) or particularly high temperatures (regeneration at 300°C or more) or using rapid temperature variations. It is understood that the sudden arrival of a hot fluid on an adsorbent at ambient temperature during the regeneration of a TSA, the massive adsorption of a constituent with a high adsorption heat or the sending of a gas to be treated at a temperature below -100°C onto particles at ambient temperature can create significant stresses within said particle, in particular between its external part and its center, stresses likely to break this particle. Systems exist for simulating these rapid changes by transfers between thermostatically controlled enclosures.This allows variations of a few hundred degrees Celsius to be obtained in about ten seconds if necessary. This same type of enclosure also allows the effect of the temperature level to be tested. The integrity of the . particles by sieving or by automatic photographic image processing software. In our example, the adsorbents remain in a very classic temperature range and here, from experience, this type of test can be omitted without taking any particular risk.

[0129] In the same way, an adsorbent can undergo pressure shocks of high amplitude and / or frequency. In order to reduce investment by reducing the volume of adsorbent required to carry out a separation, there is a strong current trend to shorten the cycle times of TSAs and PSAs. In the depressurization and repressurization steps, this results in pressure variation rates that can currently be an order of magnitude higher than they were one or two decades ago. These rates are easy to determine depending on the process chosen as well as the state of the adsorbent (regenerated / saturated) that undergoes them. These tests are generally carried out in small tanks that are successively pressurized and depressurized. The adsorbent can also be placed in a piston in cyclic movement.The integrity of the particles is checked after a certain number of successive tests, for example five hundred, with pressures representative of the process and possibly with margins on the pressure levels and / or the pressure variation rates. Care is taken to ensure that the particles tested are in the different states representative of the unit (adsorbent "new, used, regenerated, saturated"). Some adsorbents under development see their structure change slightly with pressure. It will be necessary to carefully check their behavior in rapid processes and possibly limit the pressure differences or the pressure variation rates of said pressures in order to preserve the physical integrity of these particles.

[0130] In our example, the pressure variation speeds of the TSA are higher than those usually encountered in this type of unit but remain very much lower than the speeds encountered in PSA type units. We will therefore not carry out pressure shock resistance tests either, still from experience.

[0131] It therefore remains to guard against the risks of crushing the adsorbent during operation of the separation unit. The crushing resistance test is probably both the most essential and the most delicate to interpret. In contrast to the previous characteristics, the minimum value of the crushing resistance that the particles must satisfy is not easy to determine. Indeed, the force exerted on the adsorbent during a cycle depends on numerous parameters that vary greatly from one unit to another. Taking the example of the radial circulation adsorber as shown diagrammatically in [Fig. 3], it will be noted that the adsorbent is subjected to different forces or constraints: the weight of the adsorbent located above the envisaged position, the pressure exerted by the membrane and the mass of metal balls, all complicated by silo effects due to the geometry of the adsorber, the effect of the moving fluid which can lead to significant pressure effects during depressurization or re-pressurization, the reaction of the grids in particular expansion or contraction under the effect of temperature variations. There is therefore no single limit value but it can vary by an order of magnitude between the case of a low-height cylindrical adsorber intended for a guard bed (continuous operation at constant pressure and temperature) and a large radial adsorber in which rapid TSA-type cycles are implemented.

[0132] The crushing resistance that an adsorbent must have can be estimated by calculation, in a simple way in the first case cited, and in a more complex way in the case of the radial adsorber. It should be noted that there are strain gauges or preferably very small piezoelectric sensors that can be placed at a few strategic points without disturbing the operation of the unit. This makes it possible to validate - or calibrate - the calculations relating to the forces to which the particles (and also the holding grids) are subjected in order to possibly reduce the margins retained for future units. Models of a smaller size than industrial units but representative in terms of materials (shell, grids, adsorbent), equipped with such gauges or sensors can be built in order to validate 3D models.Some of the data required for a theoretical calculation are normally known to the designer because they are essential for correctly sizing a radial type adsorber, such as the elasticity of the grids and the coefficients of thermal expansion. This same type of information is generally less known regarding the adsorbent masses. It may be necessary to carry out specific measurements internally or to call on a specialized laboratory to obtain value ranges for these physical characteristics (elasticity, thermal conductivity, expansion of a particle and especially of a dense bed of particles) for the different types of adsorbents. This gives the values ​​of the pressure stresses to which the adsorbent particles may be subjected due to the geometry of the adsorber. It is also necessary to take into account the effects of the circulating gas and gravity (weight of the adsorbent, excess weight) as indicated previously.The resulting stress varies locally (depending on the position of the particles in the adsorber) and temporally (depending on the stages of the adsorption cycle).

[0133] Considering that the adsorbents can be of different dimensions, the force exerted on the particle is generally expressed subsequently in pressure, for example in practice in N / mm2 (the force can be divided by the square of the equivalent diameter or by a characteristic section). In the case of the example, we are led to fix a minimum resistance threshold of 3.0 N / mm2 higher than the usual value rather of the order of 2 N / mm2. The value retained corresponds to the regeneration and concerns the majority of the adsorbent mass due to the silo effect which limits certain constraints such as the pressure exerted by the masses located above.

[0134] It is known that under these conditions, the zeolite selected, even if it has been used until now in slightly more favorable conditions, will not pose a problem, as it still has a sufficient margin with regard to the risks of crushing. In particular, apart from the traces of water taken into account in the tests already carried out, it is only in contact in this process with constituents which do not have a notable effect on the crushing resistance. It should nevertheless be noted that certain 13X zeolites with crushing resistances close to 2 N / mm2 would not be compatible with the unit envisaged here. On the other hand, activated alumina, being a non-referenced product, must be carefully evaluated. In particular, it is advisable to test used and saturated particles which usually lose a lot of their initial resistance as a new and regenerated product.In the case of the example, it is the humidity contained in the gas which will cause both an aging phenomenon and a reduction in crushing resistance.

[0135] Equipment for measuring the crushing strength of particles is widely available and equipped with software that automatically processes the results. For example, it allows 25 to 50 particles to be tested relatively quickly and the result to be given as a mean and standard deviation. For conventional adsorbents, a device is generally used that can measure forces between 0.5 and 50 N. For beads with a diameter ranging from 1 to 1.5 mm, the breaking force is generally in the range of 1.5 to 10 N.

[0136] In the case of activated aluminas and for cycles with moderately high temperature constraints, we generally observe, compared to the "new and regenerated" product, a degradation in the mechanical strength of the "used" product which seems to tend, depending on the aging time, towards an asymptote indicating a potential loss of 10 to 35% depending on the supplier and / or the type of product from the same supplier. To this is added almost systematically a reduction of 10 to 20% due to the presence of adsorbed water (saturated state). In the case of the example, it will be noted, taking into account the tests, that the activated alumina in service may lose around 50% of its initial mechanical strength in the inlet zone of the gas to be treated. With an initial crushing strength of 8 to 10 N / mm2, it is accepted that this adsorbent is validated for the intended application but without a large margin.It will be recommended to carry out specific monitoring of this unit (checking the settlement of the adsorbent beds, the presence of dust, the evolution of pressure losses through the adsorber, etc.).

[0137] It should be noted that it might be wise to use as a precaution a first layer of alumina less sensitive to aging but this type of arrangement is gener- too complex to be implemented in a radial adsorber. It can nevertheless be a solution for more conventional adsorbers.

[0138] It should also be noted that for activated aluminas, the trends towards a degradation of mechanical resistance on the one hand with aging and on the other hand with water content seem general, with certainly differences between the products, which justify the tests mentioned above. The results obtained with zeolites are much more dispersed whether in the tests carried out internally or in the data in the literature. It is difficult, if not impossible, to derive clear rules concerning both aging and the effect of the presence or absence of adsorbate. In such a study, binderless zeolites appear more resistant than conventional zeolites with binder while other tests lead to opposite results. In particular, significant differences are mentioned between various X zeolites studied in the context of energy storage.In the articles examined, there are no clear-cut technical explanations but different avenues to explain the observed discrepancies, such as the manufacturing process of the crystals constituting the zeolite which would make them more or less sensitive to a loss of crystallinity thus giving more fragile amorphous zones. The binder also has an important role by its nature, its percentage in relation to the active zeolite, the post-treatments which can affect the porosity, the shaping process. The distribution of the pore size is also supposed to have an effect. In particular, pores of small diameters can promote capillary condensation accentuating both the effects of aging and a reduction in mechanical resistance, highly adsorbable products causing internal stresses on the structure of the adsorbent which without bursting the latter reduce the resistance to external forces.

[0139] The TSA example described in detail above shows the approach to follow to ensure that an adsorbent that is to be used in a separation unit will retain its physical integrity for the expected duration of operation. It is appropriate to list all the scenarios that could lead to degradation of the adsorbent from its provision by the supplier until its replacement after a certain number of years. Emptying the adsorber to extract the adsorbent may be part of the scenarios since this adsorbent must then be reused. For each of these scenarios, it is appropriate to determine the stress to which the adsorbent is subjected, by calculation and / or experiment, and then to compare with that which it can withstand.When it is an adsorbent well known to the designer of the separation unit, there may only be one scenario requiring verification before validation because the separation unit has a particularity compared to the usual units, whether geometric or related to the operating parameters. On the other hand, if it is a new . adsorbent, it will be necessary to carry out as many checks as possible. In the presence of a high boiling point constituent, it will also be necessary to ensure that the latter, once adsorbed, does not significantly modify the known or recently determined mechanical resistance results.

Claims

Claims

1. Method for mechanically qualifying particles of an adsorbent (66, 67) intended for a process for separation by adsorption of a gas mixture into a most adsorbable fraction and a least adsorbable fraction, the process using at least one adsorber (11, 16,... 60), said method comprising the following steps: a) identifying a set of operating scenarios for the adsorbent (66, 67) each of which leads to mechanical stress being exerted on the particles, said set of scenarios comprising subjecting the particles to mechanical, thermal or pressure shocks, crushing the particles, setting the particles in motion by a gas flow or by handling, b) determining for each of these scenarios a maximum value of the mechanical stress exerted, as a function of the geometry of the adsorber (11, 16,... 60), its method of filling it with adsorbent (66, 67), operating parameters of the process, the operating parameters comprising a pressure, a gas flow and a temperature in the adsorber (11, 16,... 60) and as a function of a variation of these parameters, c) constituting batches of N adsorbent particles (66, 67), with N > 25, of which at least one batch consists of new adsorbent and at least one batch consists of adsorbent having been loaded with the most adsorbable fraction and regenerated repeatedly, so as to obtain an aged adsorbent, d) subjecting the batches formed in step c) to a procedure for regenerating the adsorbent (66, 67) in order to desorb any impurities, so as to obtain regenerated batches, e) selecting one of the scenarios identified in step a), f) subjecting the batches regenerated in step d) to a particle rupture test in the scenario selected in step e), so as to determine an average mechanical resistance value of the particles of each batch in the selected scenario, (g) comparing the average mechanical strength value determined in step (f) for a given batch with the maximum value of mechanical stress determined in step (b) for the scenario selected in step (e), h) validate the adsorbent (66, 67) for the separation process if the value of average mechanical resistance determined in step f) is greater than the maximum value of the mechanical stress for the scenario selected in step e), said method being characterized in that it further comprises the following additional steps: (i) measuring the quantity in the gas mixture of at least one constituent whose boiling point at atmospheric pressure is greater than -20°C, constituents selected from water, hydrocarbons, ethers, alcohols, aldehydes, ketones, j) comparing the quantity measured in step i) with a predetermined threshold value, k) if the quantity measured in step i) is greater than the threshold value, load the batches regenerated in step d) by constituting it or them, and submit the batches thus obtained to the test, comparison and validation steps corresponding to steps f), g), h).